repositories

The repository pattern in Swift separates data access logic by defining protocol-based contracts for data sources, managing DTO-to-domain model mapping, and enforcing dependency injection so ViewModels depend on abstractions rather than concrete implementations.

Your ViewModel is tightly coupled because it likely consumes raw database models directly. Introducing DTO-to-domain mapping and layer separation enforces a strict data flow—API to DTO to Repository to Domain Model—keeping UI layers independent of data sources.

The best way to structure clean architecture data flow with Supabase is enforcing a strict pipeline: Supabase API fetches DTOs, a Repository maps them to domain models, and dependency injection passes these models to ViewModels, ensuring modular and testable code.

Yes, you can use dependency injection with Swift protocols to test data access logic. By defining repository protocols and injecting concrete implementations like a Supabase repository, ViewModels remain testable and independent of live data sources.

You map data transfer objects to domain models by defining a repository that retrieves DTOs representing database schemas and transforms them into clean domain models, abstracting the data source and providing type safety for your application.

You need a repository protocol for a new Swift entity when building a feature that fetches data and requires testability. Define a protocol, implement a concrete repository against a backend like Supabase, and inject it to maintain separation of concerns.